Each day, nearly 3,700 people die in car accidents around the world; that makes driving the leading cause of death for children and young adults between the ages of 5 and 29 years old, according to the Centers for Disease Control and Prevention. Moreover, roadway injuries are expected to cost the global economy $1.8 trillion between 2015 and 2030.
While deaths and injuries are certainly the greatest negatives, driving also imposes another cost on the economy: lost productivity. Every minute you spend behind the wheel is a minute that could be spent doing something else.
By dramatically reducing roadway deaths and boosting productivity, autonomous vehicle (AV) technology has the potential to change the world. Investors looking to cash in on this trend should consider NVIDIA (NASDAQ:NVDA) and Tesla (NASDAQ:TSLA). Here's why.
NVIDIA's graphics processing units (GPUs) are regarded as the gold standard in the AI industry. As evidence, the company recently set records in every category at the MLPerf benchmarks, a series of trials designed to test AI technology. In other words, NVIDIA bested all rivals in both training (learning) and inference (decision making), the two components of artificial intelligence.
NVIDIA DRIVE is an end-to-end platform, combining hardware and software to power autonomous vehicles. The NVIDIA Orin system-on-a-chip (SoC) -- which blends NVIDIA GPUs and ARM central processing units (CPUs) to create a beast that can perform over 250 trillion operations per second (TOPS) -- will hit vehicle production lines by 2022.
Even more impressive, the chipmaker recently announced that Orin's successor, Atlan, will be available by 2025, delivering 1,000 TOPS. To put those figures in perspective, Intel-owned Mobileye's latest platform, the EyeQ5, is capable of 24 TOPS -- in other words, it's 10 times less powerful than Orin, and 40 times less powerful than Atlan. Also noteworthy, Orin delivers roughly 3 TOPS per watt, making it more power-efficient than the EyeQ5, which delivers 2.4 TOPS per watt.
NVIDIA also provides the software frameworks necessary for developers to build AV applications. When combined with NVIDIA hardware, these applications enable a vehicle to integrate sensor data (camera, radar, lidar), perceive its surroundings, and move autonomously through its environment. Not surprisingly, Navigant recognized NVIDIA's AV computing platform as the industry leader in a report published in 2020.
NVIDIA already partners with automakers like Audi, Mercedes-Benz, Nio, Toyota, and Volvo, as well as robotaxi companies like Cruise and Zoox. However, given its best-in-class solution, NVIDIA is well positioned to pick up new partners.
Going forward, NVIDIA expects its current customer pipeline to generate north of $8 billion over the next six years, but management believes its addressable market in the AV industry will reach $60 billion by 2030. That leaves a lot of upside for this semiconductor company and its shareholders.
Tesla manufactures a variety of electric vehicles, including the most recently released Model Y SUV and Model 3 sedan. Tesla began outfitting its vehicles with the latest autopilot and full self-driving hardware in October 2016: radar, eight cameras, and 12 ultrasonic sensors. That effectively turned Tesla's global fleet into a valuable source of data.
In February 2020, Tesla Director of Artificial Intelligence Andrej Karpathy provided a staggering statistic: The company had collected almost 3 billion miles of real-world driving data. By comparison, Alphabet's Waymo stated it had over 20 million miles of data in March 2020 -- 150 times less than Tesla.
Here's why that matters: Autonomous vehicles rely on powerful hardware and software to run deep neural networks (AI engines), which use input from various sensors to make decisions like when to stop or change lanes. But before these deep neural networks are capable of decision making, they must first be trained with lots of data (i.e. they need to understand when it's necessary to stop or change lanes).
Put simply, Tesla is miles ahead of other automakers in terms of data, and that gives the company a serious advantage when training its deep neural networks. But Tesla also has another advantage: Its platform is more scalable.
Waymo relies on lidar technology to map the world in high-definition; its Waymo One service then uses those maps to orchestrate the safe travel of robotaxis in Phoenix. By comparison, Tesla's approach centers on computer vision, which relies on cameras to collect data, which is then sent back to Tesla and used to improve the decision-making capabilities of its autopilot platform. This approach is much more scalable.
Ultimately, Waymo's method means its vehicles are very good at driving in Phoenix, but they would likely have great difficulty operating outside of that environment -- or if anything drastic changed within the environment, like new construction. By comparison, Tesla's use of cameras makes it platform applicable to a much wider range of geographies and scenarios.
In terms of market opportunity, Cathie Wood's Ark Invest believes autonomous ride-hailing platforms will earn over $1 trillion in annual profits by 2030. Elon Musk has said Tesla is open to licensing its autopilot platform, which would allow the company to tap that massive market. But even if that doesn't happen, Ark believes manufacturers of autonomous vehicles will reap $250 billion in annual profits by 2030. That leaves plenty of upside for Tesla and its shareholders.